Self-Driving Car Engineer Nanodegree¶
Deep Learning¶
Project: Build a Traffic Sign Recognition Classifier¶
In this notebook, a template is provided for you to implement your functionality in stages which is required to successfully complete this project. If additional code is required that cannot be included in the notebook, be sure that the Python code is successfully imported and included in your submission, if necessary. Sections that begin with 'Implementation' in the header indicate where you should begin your implementation for your project. Note that some sections of implementation are optional, and will be marked with 'Optional' in the header.
In addition to implementing code, there will be questions that you must answer which relate to the project and your implementation. Each section where you will answer a question is preceded by a 'Question' header. Carefully read each question and provide thorough answers in the following text boxes that begin with 'Answer:'. Your project submission will be evaluated based on your answers to each of the questions and the implementation you provide.
Note: Code and Markdown cells can be executed using the Shift + Enter keyboard shortcut. In addition, Markdown cells can be edited by typically double-clicking the cell to enter edit mode.
Step 0: Load The Data¶
# Load pickled data
import pickle
training_file = 'train.p'
testing_file = 'test.p'
with open(training_file, mode='rb') as f:
train = pickle.load(f)
with open(testing_file, mode='rb') as f:
test = pickle.load(f)
X_train, y_train = train['features'], train['labels']
X_test, y_test = test['features'], test['labels']
print("Data Loaded")
Step 1: Dataset Summary & Exploration¶
The pickled data is a dictionary with 4 key/value pairs:
'features'is a 4D array containing raw pixel data of the traffic sign images, (num examples, width, height, channels).'labels'is a 2D array containing the label/class id of the traffic sign. The filesignnames.csvcontains id -> name mappings for each id.'sizes'is a list containing tuples, (width, height) representing the the original width and height the image.'coords'is a list containing tuples, (x1, y1, x2, y2) representing coordinates of a bounding box around the sign in the image. THESE COORDINATES ASSUME THE ORIGINAL IMAGE. THE PICKLED DATA CONTAINS RESIZED VERSIONS (32 by 32) OF THESE IMAGES
Complete the basic data summary below.
### Replace each question mark with the appropriate value.
# Number of training examples?
n_train = len(X_train)
# Number of testing examples?
n_test = len(X_test)
# What's the shape of a traffic sign image?
image_shape = X_train[0].shape
# How many unique classes/labels are there in the dataset?
n_classes = 42
print("Number of training examples =", n_train)
print("Number of testing examples =", n_test)
print("Image data shape =", image_shape)
print("Number of classes =", n_classes)
Visualize the German Traffic Signs Dataset using the pickled file(s). This is open ended, suggestions include: plotting traffic sign images, plotting the count of each sign, etc.
The Matplotlib examples and gallery pages are a great resource for doing visualizations in Python.
NOTE: It's recommended you start with something simple first. If you wish to do more, come back to it after you've completed the rest of the sections.
### Data exploration visualization goes here.
### Feel free to use as many code cells as needed.
import random
import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline
index = random.randint(0, len(X_train))
image = X_train[index].squeeze()
plt.figure(figsize=(1,1))
plt.imshow(image)
print(y_train[index])
Step 2: Design and Test a Model Architecture¶
Design and implement a deep learning model that learns to recognize traffic signs. Train and test your model on the German Traffic Sign Dataset.
There are various aspects to consider when thinking about this problem:
- Neural network architecture
- Play around preprocessing techniques (normalization, rgb to grayscale, etc)
- Number of examples per label (some have more than others).
- Generate fake data.
Here is an example of a published baseline model on this problem. It's not required to be familiar with the approach used in the paper but, it's good practice to try to read papers like these.
NOTE: The LeNet-5 implementation shown in the classroom at the end of the CNN lesson is a solid starting point. You'll have to change the number of classes and possibly the preprocessing, but aside from that it's plug and play!
Implementation¶
Use the code cell (or multiple code cells, if necessary) to implement the first step of your project. Once you have completed your implementation and are satisfied with the results, be sure to thoroughly answer the questions that follow.
import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec
import numpy as np
%matplotlib inline
import matplotlib.image as mpimg
plt.hist(y_train,bins='auto')
plt.show()
### PREPROCESSING STEP: Convert images to YUV and then normalize contrast
import cv2
def preprocess(img_rgb):
img_yuv = cv2.cvtColor(img_rgb, cv2.COLOR_RGB2YUV)
img_yuv_normg = img_yuv
img_yuv_normg[:,:,0] = cv2.equalizeHist(img_yuv[:,:,0])
return img_yuv_normg
### PREPROCESSING STEP: Compare preprocessed images (top half of grid below) to raw images (bottom half of grid below)
gs1 = gridspec.GridSpec(10, 10)
gs1.update(wspace=0.01, hspace=0.02)
plt.figure(figsize=(12,12))
j=51
for i in range(50):
ax1 = plt.subplot(gs1[i])
ax1.set_xticklabels([])
ax1.set_yticklabels([])
ax1.set_aspect('equal')
index = random.randint(0, len(X_train))
image = X_train[index].squeeze()
img = preprocess(image)
plt.subplot(10,10,i+1)
plt.imshow(img)
plt.subplot(10,10,j)
j=j+1
plt.imshow(image)
plt.axis('off')
plt.show()
### PREPROCESSING STEP: Preprocess the entire training dataset and test datasets
# Training Data
r = len(X_train)
X_train_proc = X_train
print('X_train is size',r)
for idx in range(r):
rawimage = X_train[idx].squeeze()
procimg = preprocess(rawimage)
X_train_proc[idx] = procimg
y_train_proc = y_train
print('Preprocessing of X_train complete')
print('X_train_proc is size',len(X_train_proc))
print('Y_train_proc is size',len(y_train_proc))
### PREPROCESSING STEP: Preprocess the entire test dataset (YUV conversion and contrast normalization)
r = len(X_test)
X_test_proc = X_test
print('X_test is size',r)
for idx in range(r):
rawimage = X_test[idx].squeeze()
procimg = preprocess(rawimage)
X_test_proc[idx] = procimg
y_test_proc = y_test
print('Preprocessing of X_test complete')
print('X_test_proc is size',len(X_test_proc))
print('Y_test_proc is size',len(y_test_proc))
### PREPROCESSING STEP: Recheck the preprocessing to make sure it worked on a random sample
index = random.randint(0, len(X_train))
image = X_train_proc[index].squeeze()
plt.figure(figsize=(1,1))
plt.imshow(image)
print(y_train_proc[index])
Question 1¶
Describe how you preprocessed the data. Why did you choose that technique?
Answer:
I read in the the Sermanet and LeCun paper that they had converted RGB images to YUV format and then done some contrast normalization on the data as well as the post by Vivek Yadav on the Term 1 resources page.
### ADD ADDITIONAL DATA: Functions to create additional images via image transforms
# code from Vivek Yadav: https://github.com/vxy10/ImageAugmentation
def augment_brightness_camera_images(image):
image1 = cv2.cvtColor(image,cv2.COLOR_RGB2HSV)
random_bright = .25+np.random.uniform()
image1[:,:,2] = image1[:,:,2]*random_bright
image1 = cv2.cvtColor(image1,cv2.COLOR_HSV2RGB)
return image1
def transform_image(img,ang_range,shear_range,trans_range):
'''
This function transforms images to generate new images.
The function takes in following arguments,
1- Image
2- ang_range: Range of angles for rotation
3- shear_range: Range of values to apply affine transform to
4- trans_range: Range of values to apply translations over.
A Random uniform distribution is used to generate different parameters for transformation
'''
# Rotation
ang_rot = np.random.uniform(ang_range)-ang_range/2
rows,cols,ch = img.shape
Rot_M = cv2.getRotationMatrix2D((cols/2,rows/2),ang_rot,1)
# Translation
tr_x = trans_range*np.random.uniform()-trans_range/2
tr_y = trans_range*np.random.uniform()-trans_range/2
Trans_M = np.float32([[1,0,tr_x],[0,1,tr_y]])
# Shear
pts1 = np.float32([[5,5],[20,5],[5,20]])
pt1 = 5+shear_range*np.random.uniform()-shear_range/2
pt2 = 20+shear_range*np.random.uniform()-shear_range/2
# Brightness
pts2 = np.float32([[pt1,5],[pt2,pt1],[5,pt2]])
shear_M = cv2.getAffineTransform(pts1,pts2)
img = cv2.warpAffine(img,Rot_M,(cols,rows))
img = cv2.warpAffine(img,Trans_M,(cols,rows))
img = cv2.warpAffine(img,shear_M,(cols,rows))
#img = augment_brightness_camera_images(img)
return img
### ADDITIONAL DATA: Sample and Plot Transformed Images
gs1 = gridspec.GridSpec(10, 10)
gs1.update(wspace=0.01, hspace=0.02) # set the spacing between axes.
plt.figure(figsize=(12,12))
index = random.randint(0, len(X_train))
for i in range(100):
ax1 = plt.subplot(gs1[i])
ax1.set_xticklabels([])
ax1.set_yticklabels([])
ax1.set_aspect('equal')
image = X_train_proc[index].squeeze()
img = transform_image(image,15,5,0)
plt.subplot(10,10,i+1)
plt.imshow(img)
plt.axis('off')
plt.show()
### ADDITIONAL DATA: Add transformed images
# Training Data
r = len(X_train_proc)
num2add = 4
X_train_final = np.empty([r*num2add,32,32,3])
y_train_final = np.empty([r*num2add,1])
print('X_train_proc is size',r)
print('Generation of additional data started...')
j=0
for idx in range(r):
rawimage = X_train_proc[idx].squeeze()
for jdx in range(num2add):
procimg = transform_image(rawimage,15,5,0)
#procimg = rawimage
X_train_final[j] = procimg
y_train_final[j] = y_train[idx]
j=j+1
print('Generation of additional data complete')
print('X_train_final is size',len(X_train_final))
print('y_train_final is size',len(y_train_final))
# Split 80% of the data into training data
# Split 20% of the data into test data
from sklearn.model_selection import train_test_split
X_train_final, X_validation_data, y_train_final, y_validation_data = train_test_split(X_train_final, y_train_final, test_size=0.2, random_state=0)
print("Updated Image Shape: ",format(X_train_final[0].shape))
from sklearn.utils import shuffle
X_train_final, y_train_final = shuffle(X_train_final,y_train_final)
print('Separation of Training and Test Data Complete')
print('X_train_final is size',len(X_train_final))
print('y_train_final is size',len(y_train_final))
print('X_validation_data is size',len(X_validation_data))
print('y_validation_data is size',len(y_validation_data))
Question 2¶
Describe how you set up the training, validation and testing data for your model. Optional: If you generated additional data, how did you generate the data? Why did you generate the data? What are the differences in the new dataset (with generated data) from the original dataset?
Answer:
### Define your architecture here.
import tensorflow as tf
EPOCHS = 30
BATCH_SIZE = 128
from tensorflow.contrib.layers import flatten
def LeNet(x):
# HyperParameters
mu = 0
sigma = 0.1
# Layer 1: Convolutional. Input = 32x32x3. Output = 28x28x6.
conv1_W = tf.Variable(tf.truncated_normal(shape=(5, 5, 3, 6), mean = mu, stddev = sigma))
conv1_b = tf.Variable(tf.zeros(6))
conv1 = tf.nn.conv2d(x, conv1_W, strides=[1, 1, 1, 1], padding='VALID') + conv1_b
# Activation.
conv1 = tf.nn.relu(conv1)
# Pooling. Input = 28x28x6. Output = 14x14x6.
conv1 = tf.nn.max_pool(conv1, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='VALID')
# Layer 2: Convolutional. Output = 10x10x16.
conv2_W = tf.Variable(tf.truncated_normal(shape=(5, 5, 6, 16), mean = mu, stddev = sigma))
conv2_b = tf.Variable(tf.zeros(16))
conv2 = tf.nn.conv2d(conv1, conv2_W, strides=[1, 1, 1, 1], padding='VALID') + conv2_b
# Activation.
conv2 = tf.nn.relu(conv2)
# Pooling. Input = 10x10x16. Output = 5x5x16.
conv2 = tf.nn.max_pool(conv2, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='VALID')
# Flatten. Input = 5x5x16. Output = 400.
fc0 = flatten(conv2)
# Layer 3: Fully Connected. Input = 400. Output = 120.
fc1_W = tf.Variable(tf.truncated_normal(shape=(400, 120), mean = mu, stddev = sigma))
fc1_b = tf.Variable(tf.zeros(120))
fc1 = tf.matmul(fc0, fc1_W) + fc1_b
# Activation.
fc1 = tf.nn.relu(fc1)
# Layer 4: Fully Connected. Input = 120. Output = 84.
fc2_W = tf.Variable(tf.truncated_normal(shape=(120, 84), mean = mu, stddev = sigma))
fc2_b = tf.Variable(tf.zeros(84))
fc2 = tf.matmul(fc1, fc2_W) + fc2_b
# Activation.
fc2 = tf.nn.relu(fc2)
# Layer 5: Fully Connected. Input = 84. Output = 43.
fc3_W = tf.Variable(tf.truncated_normal(shape=(84, 43), mean = mu, stddev = sigma))
fc3_b = tf.Variable(tf.zeros(43))
logits = tf.matmul(fc2, fc3_W) + fc3_b
print(logits.get_shape())
return logits
Question 3¶
What does your final architecture look like? (Type of model, layers, sizes, connectivity, etc.) For reference on how to build a deep neural network using TensorFlow, see Deep Neural Network in TensorFlow from the classroom.
Answer:
### Train your model here.
x = tf.placeholder(tf.float32, shape=(None, 32, 32, 3))
y = tf.placeholder(tf.int32, shape=(None,))
one_hot_y = tf.one_hot(y, 43)
rate = 0.001
logits = LeNet(x)
cross_entropy = tf.nn.softmax_cross_entropy_with_logits(logits=logits, labels=one_hot_y)
loss_operation = tf.reduce_mean(cross_entropy)
optimizer = tf.train.AdamOptimizer(learning_rate = rate)
training_operation = optimizer.minimize(loss_operation)
correct_prediction = tf.equal(tf.argmax(logits, 1), tf.argmax(one_hot_y, 1))
accuracy_operation = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
saver = tf.train.Saver()
def evaluate(X_data, y_data):
num_examples = len(X_data)
total_accuracy = 0
total_loss = 0
sess = tf.get_default_session()
for offset in range(0, num_examples, BATCH_SIZE):
batch_x, batch_y = X_data[offset:offset+BATCH_SIZE], y_data[offset:offset+BATCH_SIZE]
batch_x = np.squeeze(batch_x)
batch_y = np.squeeze(batch_y)
accuracy,loss = sess.run([accuracy_operation, loss_operation], feed_dict={x: batch_x, y: batch_y})
total_accuracy += (accuracy * len(batch_x))
total_loss += (loss * len(batch_x))
yield (total_accuracy / num_examples)
yield (total_loss / num_examples)
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
num_examples = len(X_train_final)
print("Training...")
print()
for i in range(EPOCHS):
X_train_final, y_train_final = shuffle(X_train_final, y_train_final)
for offset in range(0, num_examples, BATCH_SIZE):
end = offset + BATCH_SIZE
batch_x, batch_y = np.squeeze(X_train_final[offset:end]), np.squeeze(y_train_final[offset:end])
sess.run(training_operation, feed_dict={x: batch_x, y: batch_y})
validation_accuracy,validation_loss = evaluate(X_validation_data, y_validation_data)
print("EPOCH {} ...".format(i+1))
print("Validation Accuracy = {:.3f}".format(validation_accuracy))
print("Validation Loss = {:.3f}".format(validation_loss))
print()
#if (validation_accuracy > 0.96) and (validation_loss<0.1):
# break
saver.save(sess, 'lenet.ckpt')
print("Model saved")
#test_acc,test_loss = evaluate(X_test_proc,y_test_proc)
#print("Test Accuracy = {:.3f}".format(test_acc))
accuracy = tf.reduce_mean
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
tf.train.import_meta_graph('lenet.ckpt.meta')
saver.restore(sess,tf.train.latest_checkpoint('./'))
test_acc,test_loss = evaluate(X_test_proc,y_test_proc)
print("Test Accuracy = {:.3f}".format(test_acc))
Question 4¶
How did you train your model? (Type of optimizer, batch size, epochs, hyperparameters, etc.)
Answer:
Question 5¶
What approach did you take in coming up with a solution to this problem? It may have been a process of trial and error, in which case, outline the steps you took to get to the final solution and why you chose those steps. Perhaps your solution involved an already well known implementation or architecture. In this case, discuss why you think this is suitable for the current problem.
Answer:
Step 3: Test a Model on New Images¶
Take several pictures of traffic signs that you find on the web or around you (at least five), and run them through your classifier on your computer to produce example results. The classifier might not recognize some local signs but it could prove interesting nonetheless.
You may find signnames.csv useful as it contains mappings from the class id (integer) to the actual sign name.
Implementation¶
Use the code cell (or multiple code cells, if necessary) to implement the first step of your project. Once you have completed your implementation and are satisfied with the results, be sure to thoroughly answer the questions that follow.
# I collected 2 sets of data
# The first set is 5 screen captures from Google Earth Street view in Stuttgart Germany
# The second set are 15 cellphone camera images collected at night from my car in Minnesota
### LOAD DATA SET 1
German2 = cv2.imread("TrafficSigns/German/jpg/TrafficSign2.jpg",3)
German2 = cv2.cvtColor(German2,cv2.COLOR_BGR2RGB)
German5 = cv2.imread("TrafficSigns/German/jpg/TrafficSign5.jpg",3)
German5 = cv2.cvtColor(German5,cv2.COLOR_BGR2RGB)
German6 = cv2.imread("TrafficSigns/German/jpg/TrafficSign6.jpg",3)
German6 = cv2.cvtColor(German6,cv2.COLOR_BGR2RGB)
German7 = cv2.imread("TrafficSigns/German/jpg/TrafficSign7.jpg",3)
German7 = cv2.cvtColor(German7,cv2.COLOR_BGR2RGB)
German8 = cv2.imread("TrafficSigns/German/jpg/TrafficSign8.jpg",3)
German8 = cv2.cvtColor(German8,cv2.COLOR_BGR2RGB)
X_sampletests = np.empty([5,32,32,3])
Y_sampletests = np.empty([5])
X_sampletests[0] = preprocess(German2)
Y_sampletests[0] = 13
X_sampletests[1] = preprocess(German5)
Y_sampletests[1] = 36
X_sampletests[2] = preprocess(German6)
Y_sampletests[2] = 13
X_sampletests[3] = preprocess(German7)
Y_sampletests[3] = 12
X_sampletests[4] = preprocess(German8)
Y_sampletests[4] = 12
gs2 = gridspec.GridSpec(1, 5)
gs2.update(wspace=0.01, hspace=0.02)
plt.figure(figsize=(32,32))
ax2 = plt.subplot(gs2[0])
ax2.set_xticklabels([])
ax2.set_yticklabels([])
ax2.set_aspect('equal')
plt.subplot(1,10,1)
plt.imshow(German2) # Sign Type 13 (Yield)
plt.axis('off')
plt.subplot(1,10,2)
plt.imshow(German5) # Sign Type 36 (Straight or Right)
plt.axis('off')
plt.subplot(1,10,3)
plt.imshow(German6) # Sign Type 13 (Yield)
plt.axis('off')
plt.subplot(1,10,4)
plt.imshow(German7) # Sign Type 12 (Priority Road)
plt.axis('off')
plt.subplot(1,10,5)
plt.imshow(German8) # Sign Type 12 (Priority Road)
plt.axis('off')
plt.show()
### LOAD DATA SET 2
MN1 = cv2.imread("TrafficSigns/Mine/Resized/IMG_3804.jpg",3)
MN1 = cv2.cvtColor(MN1,cv2.COLOR_BGR2RGB)
MN2 = cv2.imread("TrafficSigns/Mine/Resized/IMG_3822.jpg",3)
MN2 = cv2.cvtColor(MN2,cv2.COLOR_BGR2RGB)
MN3 = cv2.imread("TrafficSigns/Mine/Resized/IMG_3825.jpg",3)
MN3 = cv2.cvtColor(MN3,cv2.COLOR_BGR2RGB)
MN4 = cv2.imread("TrafficSigns/Mine/Resized/IMG_3842.jpg",3)
MN4 = cv2.cvtColor(MN4,cv2.COLOR_BGR2RGB)
MN5 = cv2.imread("TrafficSigns/Mine/Resized/IMG_3846.jpg",3)
MN5 = cv2.cvtColor(MN5,cv2.COLOR_BGR2RGB)
MN6 = cv2.imread("TrafficSigns/Mine/Resized/IMG_3854.jpg",3)
MN6 = cv2.cvtColor(MN6,cv2.COLOR_BGR2RGB)
MN7 = cv2.imread("TrafficSigns/Mine/Resized/IMG_3857.jpg",3)
MN7 = cv2.cvtColor(MN7,cv2.COLOR_BGR2RGB)
MN8 = cv2.imread("TrafficSigns/Mine/Resized/IMG_3859.jpg",3)
MN8 = cv2.cvtColor(MN8,cv2.COLOR_BGR2RGB)
MN9 = cv2.imread("TrafficSigns/Mine/Resized/IMG_3861.jpg",3)
MN9 = cv2.cvtColor(MN9,cv2.COLOR_BGR2RGB)
MN10 = cv2.imread("TrafficSigns/Mine/Resized/IMG_3866.jpg",3)
MN10 = cv2.cvtColor(MN10,cv2.COLOR_BGR2RGB)
MN11 = cv2.imread("TrafficSigns/Mine/Resized/IMG_3870.jpg",3)
MN11 = cv2.cvtColor(MN11,cv2.COLOR_BGR2RGB)
MN12 = cv2.imread("TrafficSigns/Mine/Resized/IMG_3875.jpg",3)
MN12 = cv2.cvtColor(MN12,cv2.COLOR_BGR2RGB)
MN13 = cv2.imread("TrafficSigns/Mine/Resized/IMG_3877.jpg",3)
MN13 = cv2.cvtColor(MN13,cv2.COLOR_BGR2RGB)
MN14 = cv2.imread("TrafficSigns/Mine/Resized/IMG_3880.jpg",3)
MN14 = cv2.cvtColor(MN14,cv2.COLOR_BGR2RGB)
MN15 = cv2.imread("TrafficSigns/Mine/Resized/IMG_3886.jpg",3)
MN15 = cv2.cvtColor(MN15,cv2.COLOR_BGR2RGB)
X_mn_sample = np.empty([15,32,32,3])
Y_mn_sample = np.empty([15])
X_mn_sample[0] = preprocess(MN1)
Y_mn_sample[0] = 0
X_mn_sample[1] = preprocess(MN2)
Y_mn_sample[1] = 0
X_mn_sample[2] = preprocess(MN3)
Y_mn_sample[2] = 0
X_mn_sample[3] = preprocess(MN4)
Y_mn_sample[3] = 0
X_mn_sample[4] = preprocess(MN5)
Y_mn_sample[4] = 0
X_mn_sample[5] = preprocess(MN6)
Y_mn_sample[5] = 0
X_mn_sample[6] = preprocess(MN7)
Y_mn_sample[6] = 0
X_mn_sample[7] = preprocess(MN8)
Y_mn_sample[7] = 0
X_mn_sample[8] = preprocess(MN9)
Y_mn_sample[8] = 0
X_mn_sample[9] = preprocess(MN10)
Y_mn_sample[9] = 0
X_mn_sample[10] = preprocess(MN11)
Y_mn_sample[10] = 0
X_mn_sample[11] = preprocess(MN12)
Y_mn_sample[11] = 0
X_mn_sample[12] = preprocess(MN13)
Y_mn_sample[12] = 0
X_mn_sample[13] = preprocess(MN14)
Y_mn_sample[13] = 0
X_mn_sample[14] = preprocess(MN15)
Y_mn_sample[14] = 0
gs2 = gridspec.GridSpec(5, 5)
gs2.update(wspace=0.01, hspace=0.02)
plt.figure(figsize=(32,32))
ax2 = plt.subplot(gs2[0])
ax2.set_xticklabels([])
ax2.set_yticklabels([])
ax2.set_aspect('equal')
plt.subplot(5,5,1)
plt.imshow(MN1) # Stop
plt.axis('off')
plt.subplot(5,5,2)
plt.imshow(MN2) # Stop Ahead
plt.axis('off')
plt.subplot(5,5,3)
plt.imshow(MN3) # Do Not Enter
plt.axis('off')
plt.subplot(5,5,4)
plt.imshow(MN4) # School Xing
plt.axis('off')
plt.subplot(5,5,5)
plt.imshow(MN5) # Converge Left
plt.axis('off')
plt.subplot(5,5,6)
plt.imshow(MN6) # Pedestrians
plt.axis('off')
plt.subplot(5,5,7)
plt.imshow(MN7) # Crosswalk
plt.axis('off')
plt.subplot(5,5,8)
plt.imshow(MN8) # Caution Deer
plt.axis('off')
plt.subplot(5,5,9)
plt.imshow(MN9) # Keep Right
plt.axis('off')
plt.subplot(5,5,10)
plt.imshow(MN10) # 2 Direction
plt.axis('off')
plt.subplot(5,5,11)
plt.imshow(MN11) # Divided Highway
plt.axis('off')
plt.subplot(5,5,12)
plt.imshow(MN12) # No Pedestrians
plt.axis('off')
plt.subplot(5,5,13)
plt.imshow(MN13) # Stop
plt.axis('off')
plt.subplot(5,5,14)
plt.imshow(MN14) # 2 Way Traffic
plt.axis('off')
plt.subplot(5,5,15)
plt.imshow(MN15) # Do Not Enter
plt.axis('off')
plt.show()
Question 6¶
Choose five candidate images of traffic signs and provide them in the report. Are there any particular qualities of the image(s) that might make classification difficult? It could be helpful to plot the images in the notebook.
Answer:
In the GERMAN SAMPLE SET:
In the USA SAMPLE SET:
### Run the predictions for the 5 German road signs from Google Street View
prediction = tf.argmax(tf.nn.softmax(logits),1)
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
tf.train.import_meta_graph('lenet.ckpt.meta')
saver.restore(sess,tf.train.latest_checkpoint('./'))
pred = sess.run(prediction,feed_dict={x:X_sampletests})
print('Prediction:',pred[0],'Truth:',Y_sampletests[0])
print('Prediction:',pred[1],'Truth:',Y_sampletests[1])
print('Prediction:',pred[2],'Truth:',Y_sampletests[2])
print('Prediction:',pred[3],'Truth:',Y_sampletests[3])
print('Prediction:',pred[4],'Truth:',Y_sampletests[4])
acc = 4/5
print('Accuracy:',acc)
### Run the predictions here for the 15 USA road signs from my car
prediction = tf.argmax(logits,1)
with tf.Session() as sess:
tf.train.import_meta_graph('lenet.ckpt.meta')
saver.restore(sess,tf.train.latest_checkpoint('./'))
pred = sess.run(prediction,feed_dict={x:X_mn_sample, y:Y_mn_sample})
print('Stop Sign, Prediction:',pred[0])
print('Stop Sign Ahead,Prediction:',pred[1])
print('Do Not Enter, Prediction:',pred[2])
print('School Xing, Prediction:',pred[3])
print('Merge Left, Prediction:',pred[4])
print('Pedestrian, Prediction:',pred[5])
print('Crosswalk, Prediction:',pred[6])
print('Deer, Prediction:',pred[7])
print('Keep Right, Prediction:',pred[8])
print('2 Direction, Prediction:',pred[9])
print('Divided Hwy, Prediction:',pred[10])
print('No Pedestrians, Prediction:',pred[11])
print('Stop, Prediction:',pred[12])
print('2 Way, Prediction:',pred[13])
print('Do Not Enter, Prediction:',pred[14])
Question 7¶
Is your model able to perform equally well on captured pictures when compared to testing on the dataset? The simplest way to do this check the accuracy of the predictions. For example, if the model predicted 1 out of 5 signs correctly, it's 20% accurate.
NOTE: You could check the accuracy manually by using signnames.csv (same directory). This file has a mapping from the class id (0-42) to the corresponding sign name. So, you could take the class id the model outputs, lookup the name in signnames.csv and see if it matches the sign from the image.
Answer:
My model was able to classify 4 of the 5 images I took from Google Street view in Germany, meaning it is 80% accurate. I manually checked the predictions versus the truth:
Image 1:Prediction: 13 Truth: 13.0 Correct Image 2:Prediction: 25 Truth: 36.0 Incorrect Image 3:Prediction: 13 Truth: 13.0 Correct Image 4:Prediction: 12 Truth: 12.0 Correct Image 5:Prediction: 12 Truth: 12.0 Correct Resulting Accuracy: 0.8
Question 8¶
Use the model's softmax probabilities to visualize the certainty of its predictions, tf.nn.top_k could prove helpful here. Which predictions is the model certain of? Uncertain? If the model was incorrect in its initial prediction, does the correct prediction appear in the top k? (k should be 5 at most)
tf.nn.top_k will return the values and indices (class ids) of the top k predictions. So if k=3, for each sign, it'll return the 3 largest probabilities (out of a possible 43) and the correspoding class ids.
Take this numpy array as an example:
# (5, 6) array
a = np.array([[ 0.24879643, 0.07032244, 0.12641572, 0.34763842, 0.07893497,
0.12789202],
[ 0.28086119, 0.27569815, 0.08594638, 0.0178669 , 0.18063401,
0.15899337],
[ 0.26076848, 0.23664738, 0.08020603, 0.07001922, 0.1134371 ,
0.23892179],
[ 0.11943333, 0.29198961, 0.02605103, 0.26234032, 0.1351348 ,
0.16505091],
[ 0.09561176, 0.34396535, 0.0643941 , 0.16240774, 0.24206137,
0.09155967]])Running it through sess.run(tf.nn.top_k(tf.constant(a), k=3)) produces:
TopKV2(values=array([[ 0.34763842, 0.24879643, 0.12789202],
[ 0.28086119, 0.27569815, 0.18063401],
[ 0.26076848, 0.23892179, 0.23664738],
[ 0.29198961, 0.26234032, 0.16505091],
[ 0.34396535, 0.24206137, 0.16240774]]), indices=array([[3, 0, 5],
[0, 1, 4],
[0, 5, 1],
[1, 3, 5],
[1, 4, 3]], dtype=int32))Looking just at the first row we get [ 0.34763842, 0.24879643, 0.12789202], you can confirm these are the 3 largest probabilities in a. You'll also notice [3, 0, 5] are the corresponding indices.
# Add an op to initialize the variables.
softmax = tf.nn.softmax(logits)
topFive = tf.nn.top_k(softmax, k=5, sorted=True, name=None)
with tf.Session() as session:
session.run(tf.global_variables_initializer())
tf.train.import_meta_graph('lenet.ckpt.meta')
saver.restore(session,tf.train.latest_checkpoint('./'))
top5_pred = session.run(topFive, feed_dict = {x: X_sampletests})
top5_pred_sftmax = top5_pred[0]
top5_pred_labels = top5_pred[1]
print(top5_pred_sftmax)
print(top5_pred_labels)
Answer:
Note: Once you have completed all of the code implementations and successfully answered each question above, you may finalize your work by exporting the iPython Notebook as an HTML document. You can do this by using the menu above and navigating to \n", "File -> Download as -> HTML (.html). Include the finished document along with this notebook as your submission.